adobe/antialiased-cnns

Issues

• HTTP Error 403: Forbidden when loading weights

  #55 opened a year ago by janehwu
  9

• If stride=1, is there a difference between BlurPool and maxpool

  #54 opened a year ago by yingmengzhiyin
  0

• my keras impl

  #10 opened 2 years ago by iperov
  18

• Why do deeper CNNs have better shift consistency?

  #52 opened 2 years ago by vadik6666
  2

• Is there any particular reason for puting Blurpool before the skip-connection layers?

  #51 opened 2 years ago by freshn
  3

• Padding size issue for small images

  #47 opened 2 years ago by DuyguSerbes
  2

• Feature Req: Separable Convolution

  #50 opened 2 years ago by torridgristle
  0

• Feature Req: Making the channel argument optional

  #49 opened 2 years ago by torridgristle
  0

• Could you please provide a 3D implementation in pytorch?

  #48 opened 3 years ago by hashemifar
  0

• Bilinear Upsampler Implementation

  #30 opened 5 years ago by sr-frost
  1

• Any plans to explore using sinc filter for downsampling?

  #46 opened 3 years ago by PeterL1n
  1

• Max-blur-pool used in text recognition model (CRNN)

  #44 opened 4 years ago by githubharald
  1

• HTTP Error 403: Forbidden when loading weights

  #45 opened 4 years ago by zakajd
  2

• ResNet parameter "pool_only=True"

  #43 opened 4 years ago by Yee-Master
  1

• Increased memory usage vs. torchvision equivalent

  #42 opened 4 years ago by nlml
  3

• 3D implementation ?

  #36 opened 5 years ago by romainVala
  4

• Adaptive version?

  #12 opened 5 years ago by Vlad0922
  2

• About Internal feature distance for shift Equivariance visualization

  #41 opened 4 years ago by re20100801
  2

• Larger strides/downsampling factors

  #37 opened 4 years ago by chanshing
  2

• Can't find Low Pass Filter Usage in Code

  #31 opened 5 years ago by humza909
  1

• can Downsample use tensorflow?

  #34 opened 5 years ago by lsjshilly
  1

• Is the blur_kernel trainable or parametric?

  #32 opened 4 years ago by Wuhaotiandl
  1

• weights connot download can you offer other links,thanks

  #4 opened 5 years ago by dongzhi0312
  5

• Depthwise convolutions

  #35 opened 5 years ago by aditya1709
  1

• RuntimeError when training resnext50_32x4d

  #33 opened 5 years ago by NightQing
  2

• Why Conv->Relu->BlurPool ?

  #29 opened 5 years ago by mrgloom
  1

• is it possible not to use the pad layer?

  #25 opened 5 years ago by zimenglan-sysu-512
  1

• What about upsampling?

  #28 opened 5 years ago by chanshing
  12

• What about tf.image.resize?

  #15 opened 5 years ago by makercob
  3

• Unable to establish SSL connection,when use command: bash weights/download_antialiased_models.sh

  #16 opened 5 years ago by LuZeking
  3

• Smooth convolutional filters instead of feature maps?

  #27 opened 5 years ago by mrgloom
  1

• Are conv2d_transpose layers shift variant, too?

  #8 opened 5 years ago by JuheonYi
  8

• Asking for Anti-antialiased U-Net codes

  #21 opened 5 years ago by zhangfb
  2

• Is AvrPool already an antialiased downsampling method?

  #26 opened 5 years ago by mrgloom
  1

• can i cite the figure of the paper directly?

  #24 opened 5 years ago by yelusaleng
  2

• Implementation of original ResNet

  #18 opened 5 years ago by jimmy15923
  1

• No strided pooling

  #22 opened 5 years ago by a462428
  1

• Some Problems on Reproducing the Results on cub200

  #19 opened 5 years ago by yuesongtian
  2

• Visualizing shift-equivariance

  #20 opened 5 years ago by elphinkuo
  2

• Confuse abount "pool_only" in resnet.py and alexnet.py

  #17 opened 5 years ago by jimmy15923
  0

• About the implementation

  #7 opened 6 years ago by boluoweifenda
  9

• is it good for autoencoder?

  #11 opened 5 years ago by iperov
  1

• FLOPS increase?

  #9 opened 5 years ago by bonlime
  2

• IPython missing from requirements.txt

  #13 opened 5 years ago by adrnswanberg
  0

• How is this approach related to Fourier Domain Convnets?

  #6 opened 5 years ago by CeaVi
  1

• About kernel_size of initial max pooing for Resnet

  #5 opened 6 years ago by sweaterr
  5

• Utility for own model require a retrained Imagenet Pretrained ?

  #3 opened 6 years ago by JarveeLee
  5

• Use buffers instead of parameters in the Downsample layer

  #1 opened 6 years ago by sarlinpe
  2